Phospholipases A.sub.2 (PLA.sub.2) are a family of enzymes that hydrolyze the sn-2 ester linkage of membrane phospholipids resulting in release of a free fatty acid and a lysophospholipid (see, Dennis, E. A., The Enzymes, Vol. 16, pp. 307-353, Boyer, P. D., ed., Academic Press, New York, 1983). Elevated levels of type II PLA.sub.2 are correlated with a number of human inflammatory diseases. The PLA.sub.2 -catalyzed reaction is the rate-limiting step in the release of a number of pro-inflammatory mediators. Arachidonic acid, a fatty acid commonly linked at the sn-2 position, serves as a precursor to leukotrienes, prostaglandins, lipoxins and thromboxanes. The lysophospholipid can be a precursor to platelet-activating factor. PLA.sub.2 is regulated by pro-inflammatory cytokines and, thus, occupies a central position in the inflammatory cascade (see, e.g., Dennis, ibid.; Glaser, et al., TiPs Reviews 1992, 14, 92; and Pruzanski, et al., Inflammation 1992, 16, 451).
All mammalian tissues evaluated thus far have exhibited PLA.sub.2 activity. At least three different types of PLA.sub.2 are found in humans: pancreatic (type I), synovial fluid (type II) and cytosolic. Studies suggest that additional isoenzymes exist. Type I and type II, the secreted forms of PLA.sub.2, share strong similarity with phospholipases isolated from the venom of snakes. The PLA.sub.2 enzymes are important for normal functions including digestion, cellular membrane remodeling and repair, and in mediation of the inflammatory response. Both cytosolic and type II enzymes are of interest as therapeutic targets. Increased levels of the type II PLA.sub.2 are correlated with a variety of inflammatory disorders including rheumatoid arthritis, osteoarthritis, inflammatory bowel disease and septic shock, suggesting that inhibitors of this enzyme would have therapeutic utility. Additional support for a role of PLA.sub.2 in promoting the pathophysiology observed in certain chronic inflammatory disorders was the observation that injection of type II PLA.sub.2 into the footpad of rats (Vishwanath, et al., Inflammation 1988, 12, 549) or into the articular space of rabbits (Bomalaski, et al., J. Immunol. 1991, 146, 3904) produced an inflammatory response. When the protein was denatured before injection, no inflammatory response was produced.
The type II PLA.sub.2 enzyme from synovial fluid is a relatively small molecule (about 14 kD) and can be distinguished from type I enzymes (e.g., pancreatic) by the sequence and pattern of its disulfide bonds. Both types of enzymes require calcium for activity. The crystal structures of secreted PLA.sub.2 enzymes from venom and pancreatic PLA.sub.2, with and without inhibitors, have been reported (Scott, et al., Science 1990, 250, 1541). Recently, the crystal structure of PLA.sub.2 from human synovial fluid has been solved (Wery, et al., Nature 1991, 352, 79). The structures clarify the role of calcium and amino acid residues in catalysis. The calcium acts as a Lewis acid to activate the scissile ester carbonyl and bind the lipid, and a His-Asp side chain dyad acts as general base catalyst to activate a water molecule nucleophile. This is consistent with the absence of any acyl enzyme intermediates, and is also comparable to the catalytic mechanism of serine proteases. The catalytic residues and the calcium ion are at the end of a deep cleft (ca. 14 .ANG.) in the enzyme. The walls of this cleft contact the hydrocarbon portion of the phospholipid and are composed of hydrophobic and aromatic residues. The positively-charged amino-terminal helix is situated above the opening of the hydrophobic cleft. Several lines of evidence suggest that the N-terminal portion is the interfacial binding site. (see, e.g., Achari, et al., Cold Spring Harbor Symp. Quant. Biol. 1987, 52, 441; Cho, et al., J. Biol. Chem. 1988, 263, 11237; Yang, et al., Biochem. J. 1989, 262, 855; and Noel, et al., J. Am. Chem. Soc. 1990, 112, 3704).
Much work has been reported in recent years on the study of the mechanism and properties of PLA.sub.2 -catalyzed hydrolysis of phospholipids. In in vitro assays, PLA.sub.2 displays a lag phase during which the enzyme adsorbs to the substrate bilayer and a process called interfacial activation occurs. This activation may involve desolvation of the enzyme/lipid interface or a change in the physical state of the lipid around the cleft opening. The evidence favoring this hypothesis comes from studies revealing that rapid changes in PLA.sub.2 activity occur concurrently with changes in the fluorescence of a membrane probe (Burack, et al., Biochemistry 1993, 32, 583). This suggests that lipid rearrangement is occurring during the interfacial activation process. PLA.sub.2 activity is maximal around the melting temperature of the lipid, where regions of gel and liquid-crystalline lipid coexist. This is also consistent with the sensitivity of PLA2 activity to temperature and to the composition of the substrate, both of which can lead to structurally distinct lipid arrangements separated by a boundary region. Fluorescence microscopy was used to simultaneously identify the physical state of the lipid and the position of the enzyme during catalysis (Grainger, et al., FEBS Lett. 1989, 252, 73). These studies clearly show that PLA.sub.2 binds exclusively at the boundary region between liquid and solid phase lipid.
While the hydrolysis of the secondary ester bond of 1,2-diacylglycerophospholipids catalyzed by the enzyme is relatively simple, the mechanistic and kinetic picture is clouded by the complexity of the enzyme-substrate interaction. A remarkable characteristic of PLA.sub.2 is that maximal catalytic activity is observed on substrate that is aggregated (i.e., phospholipid above its critical micelle concentration), while low levels of activity are observed on monomeric substrate. As a result, competitive inhibitors of PLA.sub.2 either have a high affinity for the active site of the enzyme before it binds to the substrate bilayer or partition into the membrane and compete for the active site with the phospholipid substrate. Although a number of inhibitors appear to show promising inhibition of PLA.sub.2 in biochemical assays (see, e.g., Yuan, et al., J. Am. Chem. Soc. 1987, 109, 8071; Lombardo, et al., J. Biol. Chem. 1985, 260, 7234; Washburn, et al., J. Biol. Chem. 30 1991, 266, 5042; Campbell, et al., J. Chem. Soc., Chem. Commun. 1988, 1560; and Davidson, et al., Biochem. Biophys. Res. Commun. 1986, 137, 587), reports describing in vivo activity are limited (see, e.g., Miyake, et al., J. Pharmacol. Exp. Ther. 1992, 263, 1302).
Traditional structure activity relationship type drug discovery gives unambiguous products but yet requires the preparation of numerous individual test candidates. The preparation of each structure requires significant amounts of time and resources. Another drug discovery approach, de novo design of active compounds based on high resolution enzyme structures, generally has not been successful. Yet another approach involves screening complex fermentation broths and plant extracts for a desired biological activity. The advantage of screening mixtures from biological sources is that a large number of compounds can be screened simultaneously, in some cases leading to the discovery of novel and complex natural products with activity that could not have been predicted otherwise. One disadvantage is that many different samples must be screened and numerous purifications must be carried out to identify the active component, which often is present only in trace amounts.
In order to maximize the advantages of each classical approach, new strategies for combinatorial unrandomization have been developed by several groups. Selection techniques have been used with libraries of peptides (see, e.g., Geysen, et al., J. Immun. Meth. 1987, 102, 259; Houghten, et al., Nature 1991, 354, 84; and Owens, et al., Biochem. Biophys. Res. Commun. 1991, 181, 402) and nucleic acids (see, e.g,, Wyatt, et al., (in press) Proc. Natl. Acad. Sci. USA; and Ecker, et al., Nucleic Acids Res. 1993, 21, 1853). These selection techniques involve iterative synthesis and screening of increasingly simplified subsets of oligomers. In using these selection techniques, subsets are assayed for activity in either cell-based assays, or for binding or inhibition of purified protein targets.
One technique, called SURF (Synthetic Unrandomization of Randomized Fragments; see, e.g., Ecker, et al., ibid., involves the synthesis of subsets of oligomers containing a known residue at one fixed monomer position and equimolar mixtures of residues at all other positions. For a library of oligomers four residues long containing three monomers (A, B, C), three subsets would be synthesized (NNAN, NNBN, NNCN, where N represents equal incorporation of each of the three monomers). Each subset is then screened in a functional assay and the best subset is identified (e.g., NNAN) A second set of libraries is synthesized and screened, each containing the fixed residue from the previous round, and a second fixed residue (e.g. ANAN, BNAN, CNAN). Through successive rounds of screening and synthesis, a unique sequence with activity in the assay can be identified.